Computational Methods and Experimental Measurements XV 671 The rescheduling system for trains, drivers and crew of high speed railways H. Shimizu1, H. Tanabe2 & M. Matsumoto2 1JR East Japan Information Systems Company, Japan 2East Japan Railway Company, Japan Abstract We are developing the new rescheduling system for drivers and crew that synchronizes the train forecasted plan. The East Japan Railway Company has five Bullet Train lines (SHINKANSEN) operating about 800 trains per day, and dispatching nearly 250 drivers and 350 crew per day in order to operate the trains. To keep the operation plan and management of SHINKANSEN properly, we have the system named COSMOS (Computerized Safety Maintenance and Operation systems of SHINKANSEN). The drivers and crew rostering schedule are made with the drivers and crew rescheduling system that is one of the sub- systems of COSMOS Transportation Planning System. Each driver and crew works on their trains according to a rostering scheduling decided upon beforehand normally. However, sometimes trains would be unable to operate on schedule when the weather is bad or car troubles happen. In such cases, the rostering scheduling must be redesigned. The influence reaches the two or more crews’ rostering scheduling by changing only one crew member. It is very difficult to make this rescheduling because a delay of the train changes minute by minute. To make the crews’ rostering scheduling change adequately, we have developed a new system. This system displays some contradictions of the rescheduled plan of drivers and crew if a train delay happened. Dispatchers reschedule the plan to solve these contradictions. We reduce the delay time of trains by utilizing this system and will improve our services for customers. Keywords: train traffic rescheduling, drivers rostering rescheduling, crew rostering rescheduling, forecast, real-time rescheduling. WIT Transactions on Modelling and Simulation, Vol 51, © 2011 WIT Press www.witpress.com, ISSN 1743-355X (on-line) doi:10.2495/CMEM110591 672 Computational Methods and Experimental Measurements XV 1 Introduction East Japan Railway Company (JR-East) has five Bullet Train (SHINKANSEN) lines – Tohoku line, Joetsu line, Hokuriku line, Yamagata line, and the Akita line. Our network of SHINKANSEN is shown in Figure 1, and features of each line are shown in Table 1. These lines all start from Tokyo and directly connect to five areas in east Japan. A part of the section of the Yamagata line – between Fukushima and Shinjo and a part of section of the Akita line – between Morioka and Akita are low speed sections. In these sections, SHINKANSEN trains and local trains are operated in coexistence. Trains of Yamagata and Akita line are combined with the train of Tohoku line between Tokyo and Fukushima or between Tokyo Morioka. To respond to a variety of passenger needs in the SHINKANSEN transportation, JR-East has a variety of types of cars. The types of vehicles in April, 2010 are shown in Table 2. The transportation scale, the number of trains per day, car rostering, driver rostering and crew rostering are shown in Table 3. Akita Line Yamagata Line Joetsu Line Tohoku Line Hokuriku Line Figure 1: Network of JR-East SHINKANSEN [3]. WIT Transactions on Modelling and Simulation, Vol 51, © 2011 WIT Press www.witpress.com, ISSN 1743-355X (on-line) Computational Methods and Experimental Measurements XV 673 Table 1: Features of JR-East SHINKANSEN (As of March 2010). Line name Distance of line Name of trains Number a day of trains Tohoku Line 631.9 km Yamabiko 89 (Tokyo – Hachinohe) Hayate 37 Nasuno 38 Joetsu Line 333.9 km Toki 56 (Tokyo – Niigata) Tanigawa 39 Hokuriku Line 222.4 km Asama 56 (Tokyo – Nagano) Yamagata Line 421.4 km Tsubasa 33 (Tokyo – Shinjo) Akita Line 662.6 km Komachi 32 (Tokyo – Akita) Table 2: Kinds of vehicles of JR-East SHINKANSEN (As of March 2010). Type Name of Line chiefly Car type Partner of vehicle operated combining 200 K Joetsu 10 cars, Flat None E1 M Joetsu 12 cars, Duplex Enable E2 J Tohoku, 10 cars, Flat R Joetsu N Hokuriku 8 cars, Flat Enable E3 R Akita 6 cars, Flat J LR Yamagata 7 cars, Flat P E4 P Tohoku, 8 cars, Duplex P, LR Joetsu E5 Unsigned Tohoku 10 cars, Flat Undecided Table 3: The transportation scale of JR-East SHINKANSEN. Trains About 1,000 Car rostering About 200 Driver rostering About 250 Crew rostering About 350 2 The transportation feature in JR-East SHINKANSEN There are following three large transportation features in JR-East SHINKANSEN. WIT Transactions on Modelling and Simulation, Vol 51, © 2011 WIT Press www.witpress.com, ISSN 1743-355X (on-line) 674 Computational Methods and Experimental Measurements XV (1) All trains of five lines and share limited track capacity between Tokyo and Oomiya. (2) To operate many trains five lines in limited track capacity, some of them are combined with another. (3) There is a difference at the transportation stability level between SHINKANSEN line and conventional line, and a conventional line train delay might influence SHINKANSEN trains. There is the possibility of influencing all areas and all trains even if a delay occurs in a small area. Therefore, we need to reschedule the plan of train operation as soon as possible. For example, in conventional sections between Fukushima and Yamagata of Yamagata line, a delay of trains and operation suspension of trains occur owing to rainfall or snowfall and so on. If a delay of trains of Yamagata occurs, it influences combined trains of Tohoku. In this case, we reschedule a splitting and combining plan of trains of Yamagata line and Tohoku line. It is a simple solution that all splitting and combining plans are cancelled and each train is operated by itself. But there are the following problems. (1) An unscheduled driver and crew is needed. (2) It is difficult that we add trains to the section between Tokyo and Omiya limited line capacity. To avoid these problems, we change combined partner trains one by one. It is called “DAN-OCHI” in Japanese. One of the examples of this rescheduling is shown in figure 2. P P P P 0 0 0 0 0 0 0 0 1 2 1 2 L 0 5 2 L P L L 0 0 5 0 0 0 1 5 5 1 P 1 2 0 0 2 P P P 0 0 0 0 1 0 0 1 2 Figure 2: One of examples of combining plan rescheduling. 3 COSMOS overview 3.1 General overview In an operation of JR-East SHINKANSEN, we need to reschedule the train operation plan as soon as possible. Therefore, we have some systems called “COSMOS” for supporting our judgement of rescheduling. COSMOS has eight systems – Transportation Plan System, Operation Control System, Rolling Stock Control System, Facility Control System, Maintenance Work Control System, WIT Transactions on Modelling and Simulation, Vol 51, © 2011 WIT Press www.witpress.com, ISSN 1743-355X (on-line) Computational Methods and Experimental Measurements XV 675 Railway Yard Work Management System, Facility Monitoring System and Electric Power Control System. We are highly managing and controlling information of transportation of SHINKANSEN by cooperatively using these eight systems. Systems overview is shown in Figure 3. In this thesis, the sub-system that the drivers and crew use in the rescheduling system which is one of sub-systems of the Transportation Plan System, cooperates with the train traffic rescheduling system which is one of sub-systems of the Operation Control System is described. The overview of two systems is shown in Figure 4. This development overview of two systems was reported in our thesis [1]. This drivers and crew rescheduling system came into practical use beginning in March, 2009. Figure 3: Overview of COSMOS systems. Train forecasted diagram data forecasted Train Figure 4: Cooperation between rescheduling functions. 3.2 Effect of improvement We remade and began to use the COSMOS Transportation Planning System and Operation Control System in May, 2008. In this renewal, we redesigned some WIT Transactions on Modelling and Simulation, Vol 51, © 2011 WIT Press www.witpress.com, ISSN 1743-355X (on-line) 676 Computational Methods and Experimental Measurements XV input window and input flow of train traffic rescheduling system. Consequently, we achieved the effect shown in Table 4. The average input time has shortened two minutes or more before and after the function improvement. Because it is necessary to input a lot of train traffic rescheduling, an effect of this time crunch is large. Table 4: The average input time. Number of input times Average input time Before After Before After Window 1 2 1 1min 38sec 54sec (Run alone) Window 2 (Change 4 2 2min 36sec 1min 52sec combined partner) Window 3 (Cancel to 7 3 4min 37sec 2min 27sec combine) Example input of rescheduling 8 4 5min 52sec 3min 40sec shown in Figure 2 4 Case study 4.1 The train traffic rescheduling system This chapter introduces the train traffic rescheduling system which is one of the sub-systems of the COSMOS Operation Control System. 福 島 New combined train number Delayed train number 104B 7102B P 002 L 051 Figure 5: Input window of system. WIT Transactions on Modelling and Simulation, Vol 51, © 2011 WIT Press www.witpress.com, ISSN 1743-355X (on-line) Computational Methods and Experimental Measurements XV 677 102M 102B Delay Figure 6: Train forecasted diagram before rescheduling. 102M 104M 104B 7104B 102B Figure 7: Train forecasted diagram after rescheduling. WIT Transactions on Modelling and Simulation, Vol 51, © 2011 WIT Press www.witpress.com, ISSN 1743-355X (on-line) 678 Computational Methods and Experimental Measurements XV If a delay occurs in a conventional section, we need to reschedule the train operation plan as soon as possible.